Yun-Da Tsai

Mingjie Liu, Yun-Da Tsai, Wenfei Zhou et al.

Despite the significant progress made in code generation with large language models, challenges persist, especially with hardware description languages such as Verilog. This paper first presents an analysis of fine-tuned LLMs on Verilog coding, with synthetic data from prior methods. We identify two main issues: difficulties in handling non-textual representations (Karnaugh maps, state-transition diagrams and waveforms) and significant variability during training with models randomly making "minor" mistakes. To address these limitations, we enhance data curation by creating correct-by-construction data targeting non-textual representations. Additionally, we introduce an automated framework that generates error reports from various model checkpoints and injects these errors into open-source code to create targeted code repair data. Our fine-tuned Starcoder2-15B outperforms prior state-of-the-art results by 3.8%, 10.9%, 6.6% for pass@1 on VerilogEval-Machine, VerilogEval-Human, and RTLLM.

Pei-Fu Guo, Yun-Da Tsai, Shou-De Lin

Prompt optimization algorithms for Large Language Models (LLMs) excel in multi-step reasoning but still lack effective uncertainty estimation. This paper introduces a benchmark dataset to evaluate uncertainty metrics, focusing on Answer, Correctness, Aleatoric, and Epistemic Uncertainty. Through analysis of models like GPT-3.5-Turbo and Meta-Llama-3.1-8B-Instruct, we show that current metrics align more with Answer Uncertainty, which reflects output confidence and diversity, rather than Correctness Uncertainty, highlighting the need for improved metrics that are optimization-objective-aware to better guide prompt optimization. Our code and dataset are available at https://github.com/0Frett/PO-Uncertainty-Benchmarking.

Pei-Fu Guo, Ying-Hsuan Chen, Yun-Da Tsai et al.

In this work, we conduct an assessment of the optimization capabilities of LLMs across various tasks and data sizes. Each of these tasks corresponds to unique optimization domains, and LLMs are required to execute these tasks with interactive prompting. That is, in each optimization step, the LLM generates new solutions from the past generated solutions with their values, and then the new solutions are evaluated and considered in the next optimization step. Additionally, we introduce three distinct metrics for a comprehensive assessment of task performance from various perspectives. These metrics offer the advantage of being applicable for evaluating LLM performance across a broad spectrum of optimization tasks and are less sensitive to variations in test samples. By applying these metrics, we observe that LLMs exhibit strong optimization capabilities when dealing with small-sized samples. However, their performance is significantly influenced by factors like data size and values, underscoring the importance of further research in the domain of optimization tasks for LLMs.

Yun-Da Tsai, Mingjie Liu, Haoxing Ren

Recent work targeting large language models (LLMs) for code generation demonstrated that increasing the amount of training data through synthetic code generation often leads to exceptional performance. In this paper we explore data pruning methods aimed at enhancing the efficiency of model training specifically for code LLMs. We present techniques that integrate various clustering and pruning metrics to selectively reduce training data without compromising the accuracy and functionality of the generated code. We observe significant redundancies in synthetic training data generation, where our experiments demonstrate that benchmark performance can be largely preserved by training on only 10% of the data. Moreover, we observe consistent improvements in benchmark results through moderate pruning of the training data. Our experiments show that these pruning strategies not only reduce the computational resources needed but also enhance the overall quality code generation.

Yun-Da Tsai, Tzu-Hsien Tsai, Shou-De Lin

This paper targets a variant of the stochastic multi-armed bandit problem called good arm identification (GAI). GAI is a pure-exploration bandit problem with the goal to output as many good arms using as few samples as possible, where a good arm is defined as an arm whose expected reward is greater than a given threshold. In this work, we propose DGAI - a differentiable good arm identification algorithm to improve the sample complexity of the state-of-the-art HDoC algorithm in a data-driven fashion. We also showed that the DGAI can further boost the performance of a general multi-arm bandit (MAB) problem given a threshold as a prior knowledge to the arm set. Extensive experiments confirm that our algorithm outperform the baseline algorithms significantly in both synthetic and real world datasets for both GAI and MAB tasks.

Pei-Fu Guo, Ya-An Tsai, Chun-Chia Hsu et al.

While most reading comprehension benchmarks for LLMs focus on factual information that can be answered by localizing specific textual evidence, many real-world tasks require understanding distributional information, such as population-level trends and preferences expressed across collections of text. We introduce Text2DistBench, a reading comprehension benchmark for evaluating LLMs' ability to infer distributional knowledge from natural language. Built from real-world YouTube comments about movie and music entities, the benchmark provides models with entity metadata and associated comments, and requires them to answer distributional questions, such as estimating the proportions of positive and negative comments, or identifying the most and second most frequent topics discussed among viewers. To support reliable and long-term evaluation, the construction pipeline of Text2DistBench is fully automated and continuously updated to incorporate newly emerging entities over time. Experiments across multiple LLMs show that while models substantially outperform random baselines, performance varies widely across different distribution types and characteristics. These findings highlight both the capabilities and limitations of current LLMs in distributional reading comprehension and demonstrate the value of Text2DistBench as a practical and scalable testbed for future research.

Pei-Fu Guo, Yun-Da Tsai, Chun-Chia Hsu et al.

Evaluating cross-lingual knowledge transfer in large language models is challenging, as correct answers in a target language may arise either from genuine transfer or from prior exposure during pre-training. We present LiveCLKTBench, an automated generation pipeline specifically designed to isolate and measure cross-lingual knowledge transfer. Our pipeline identifies self-contained, time-sensitive knowledge entities from real-world domains, filters them based on temporal occurrence, and verifies them against the model's knowledge. The documents of these valid entities are then used to generate factual questions, which are translated into multiple languages to evaluate transferability across linguistic boundaries. Using LiveCLKTBench, we evaluate several LLMs across five languages and observe that cross-lingual transfer is strongly influenced by linguistic distance and often asymmetric across language directions. While larger models improve transfer, the gains diminish with scale and vary across domains. These findings provide new insights into multilingual transfer and demonstrate the value of LiveCLKTBench as a reliable benchmark for future research.

Yun-Da Tsai, Ting-Yu Yen, Keng-Te Liao et al.

Converting different modalities into generalized text, which then serves as input prompts for large language models (LLMs), is a common approach for aligning multimodal models, particularly when pairwise data is limited. Text-centric alignment method leverages the unique properties of text as a modality space, transforming diverse inputs into a unified textual representation, thereby enabling downstream models to effectively interpret various modal inputs. This study evaluates the quality and robustness of multimodal representations in the face of noise imperfections, dynamic input order permutations, and missing modalities, revealing that current text-centric alignment methods can compromise downstream robustness. To address this issue, we propose a new text-centric adversarial training approach that significantly enhances robustness compared to traditional robust training methods and pre-trained multimodal foundation models. Our findings underscore the potential of this approach to improve the robustness and adaptability of multimodal representations, offering a promising solution for dynamic and real-world applications.

Ting-Yu Yen, Yun-Da Tsai, Keng-Te Liao et al.

Converting different modalities into general text, serving as input prompts for large language models (LLMs), is a common method to align multimodal models when there is limited pairwise data. This text-centric approach leverages the unique properties of text as a modality space, transforming diverse inputs into a unified textual representation. This enables downstream models to effectively interpret various modal inputs. This study assesses the quality and robustness of multimodal representations in the presence of missing entries, noise, or absent modalities, revealing that current text-centric alignment methods compromise downstream robustness. To address this issue, we propose a new text-centric approach that achieves superior robustness compared to previous methods across various modalities in different settings. Our findings highlight the potential of this approach to enhance the robustness and adaptability of multimodal representations, offering a promising solution for dynamic and real-world applications.

Felix Liawi, Yun-Da Tsai, Guan-Lun Lu et al.

Scene Text Editing (STE) aims to substitute text in an image with new desired text while preserving the background and styles of the original text. However, present techniques present a notable challenge in the generation of edited text images that exhibit a high degree of clarity and legibility. This challenge primarily stems from the inherent diversity found within various text types and the intricate textures of complex backgrounds. To address this challenge, this paper introduces a three-stage framework for transferring texts across text images. Initially, we introduce a text-swapping network that seamlessly substitutes the original text with the desired replacement. Subsequently, we incorporate a background inpainting network into our framework. This specialized network is designed to skillfully reconstruct background images, effectively addressing the voids left after the removal of the original text. This process meticulously preserves visual harmony and coherence in the background. Ultimately, the synthesis of outcomes from the text-swapping network and the background inpainting network is achieved through a fusion network, culminating in the creation of the meticulously edited final image. A demo video is included in the supplementary material.

Nathaniel Pinckney, Chenhui Deng, Chia-Tung Ho et al.

We present the Comprehensive Verilog Design Problems (CVDP) benchmark, a new dataset and infrastructure to advance LLM and agent research in hardware design and verification. CVDP includes 783 problems across 13 task categories, covering RTL generation, verification, debugging, specification alignment, and technical Q&A authored by experienced hardware engineers. Problems are offered in both non-agentic and agentic formats. The benchmark introduces more realistic and challenging contexts than prior work, with state-of-the-art models achieving no more than 34% pass@1 on code generation. Agentic tasks$\unicode{x2013}$especially those involving RTL reuse and verification$\unicode{x2013}$are particularly difficult. Evaluation uses open-source tools and model scoring infrastructure, with comprehension tasks assessed via BLEU and LLM-based judging. CVDP reveals substantial gaps in current model capabilities, underscoring the need for continued research toward robust, real-world hardware design automation.

Yun-Da Tsai, Ting-Yu Yen, Pei-Fu Guo et al.

This research paper addresses the challenge of modality mismatch in multimodal learning, where the modalities available during inference differ from those available at training. We propose the Text-centric Alignment for Multi-Modality Learning (TAMML) approach, an innovative method that utilizes Large Language Models (LLMs) with in-context learning and foundation models to enhance the generalizability of multimodal systems under these conditions. By leveraging the unique properties of text as a unified semantic space, TAMML demonstrates significant improvements in handling unseen, diverse, and unpredictable modality combinations. TAMML not only adapts to varying modalities but also maintains robust performance, showcasing the potential of foundation models in overcoming the limitations of traditional fixed-modality frameworks in embedding representations. This study contributes to the field by offering a flexible, effective solution for real-world applications where modality availability is dynamic and uncertain.

Tzu-Hsien Tsai, Yun-Da Tsai, Shou-De Lin

Good arm identification (GAI) is a pure-exploration bandit problem in which a single learner outputs an arm as soon as it is identified as a good arm. A good arm is defined as an arm with an expected reward greater than or equal to a given threshold. This paper focuses on the GAI problem under a small threshold gap, which refers to the distance between the expected rewards of arms and the given threshold. We propose a new algorithm called lil'HDoC to significantly improve the total sample complexity of the HDoC algorithm. We demonstrate that the sample complexity of the first $λ$ output arm in lil'HDoC is bounded by the original HDoC algorithm, except for one negligible term, when the distance between the expected reward and threshold is small. Extensive experiments confirm that our algorithm outperforms the state-of-the-art algorithms in both synthetic and real-world datasets.

Chenhui Deng, Yun-Da Tsai, Guan-Ting Liu et al.

Recent advances in large language models (LLMs) have enabled near-human performance on software coding benchmarks, but their effectiveness in RTL code generation remains limited due to the scarcity of high-quality training data. While prior efforts have fine-tuned LLMs for RTL tasks, they do not fundamentally overcome the data bottleneck and lack support for test-time scaling due to their non-reasoning nature. In this work, we introduce ScaleRTL, the first reasoning LLM for RTL coding that scales up both high-quality reasoning data and test-time compute. Specifically, we curate a diverse set of long chain-of-thought reasoning traces averaging 56K tokens each, resulting in a dataset of 3.5B tokens that captures rich RTL knowledge. Fine-tuning a general-purpose reasoning model on this corpus yields ScaleRTL that is capable of deep RTL reasoning. Subsequently, we further enhance the performance of ScaleRTL through a novel test-time scaling strategy that extends the reasoning process via iteratively reflecting on and self-correcting previous reasoning steps. Experimental results show that ScaleRTL achieves state-of-the-art performance on VerilogEval and RTLLM, outperforming 18 competitive baselines by up to 18.4% on VerilogEval and 12.7% on RTLLM.

Yun-Ang Wu, Yun-Da Tsai, Shou-De Lin

In this study, we delve into the Thresholding Linear Bandit (TLB) problem, a nuanced domain within stochastic Multi-Armed Bandit (MAB) problems, focusing on maximizing decision accuracy against a linearly defined threshold under resource constraints. We present LinearAPT, a novel algorithm designed for the fixed budget setting of TLB, providing an efficient solution to optimize sequential decision-making. This algorithm not only offers a theoretical upper bound for estimated loss but also showcases robust performance on both synthetic and real-world datasets. Our contributions highlight the adaptability, simplicity, and computational efficiency of LinearAPT, making it a valuable addition to the toolkit for addressing complex sequential decision-making challenges.

Yun-Da Tsai

Large Language Models (LLMs) have achieved remarkable success across a wide range of natural language tasks, and recent efforts have sought to extend their capabilities to multimodal domains and resource-constrained environments. However, existing approaches often rely on costly supervised fine-tuning or assume fixed training conditions, limiting their generalization when facing unseen modalities, limited data, or restricted compute resources. This dissertation presents a systematic study toward generalizing LLM usability under real-world constraints. First, it introduces a robust text-centric alignment framework that enables LLMs to seamlessly integrate diverse modalities-including text, images, tables, and any modalities - via natural language interfaces. This approach supports in-context adaptation to unseen or dynamically changing modalities without requiring retraining. To enhance robustness against noisy and missing modalities, an adversarial prompting technique is proposed, generating semantically challenging perturbations at the prompt level to stress-test model reliability. Beyond multimodal setting, the dissertation investigates inference-time optimization strategies for LLMs, leveraging prompt search and uncertainty quantification to improve performance without additional model training. This perspective offers an efficient alternative to scaling model parameters or retraining from scratch. Additionally, the work addresses low-resource domains such as Verilog code generation by designing correct-by-construction synthetic data pipelines and logic-enhanced reasoning models, achieving state-of-the-art performance with minimal data. Together, these contributions form a unified effort to enhance the adaptability, scalability, and efficiency of large language models under practical constraints.

Pei-Fu Guo, Yun-Da Tsai, Shou-De Lin

Large language models (LLMs) often generate fluent but factually incorrect outputs, known as hallucinations, which undermine their reliability in real-world applications. While uncertainty estimation has emerged as a promising strategy for detecting such errors, current metrics offer limited interpretability and lack clarity about the types of uncertainty they capture. In this paper, we present a systematic framework for decomposing LLM uncertainty into four distinct sources, inspired by previous research. We develop a source-specific estimation pipeline to quantify these uncertainty types and evaluate how existing metrics relate to each source across tasks and models. Our results show that metrics, task, and model exhibit systematic variation in uncertainty characteristic. Building on this, we propose a method for task specific metric/model selection guided by the alignment or divergence between their uncertainty characteristics and that of a given task. Our experiments across datasets and models demonstrate that our uncertainty-aware selection strategy consistently outperforms baseline strategies, helping us select appropriate models or uncertainty metrics, and contributing to more reliable and efficient deployment in uncertainty estimation.

Yun-Da Tsai, Chang-Yu Chao, Liang-Yeh Shen et al.

Modern chip physical design relies heavily on Electronic Design Automation (EDA) tools, which often struggle to provide interpretable feedback or actionable guidance for improving routing congestion. In this work, we introduce a Multimodal Large Language Model Assistant (MLLMA) that bridges this gap by not only predicting congestion but also delivering human-interpretable design suggestions. Our method combines automated feature generation through MLLM-guided genetic prompting with an interpretable preference learning framework that models congestion-relevant tradeoffs across visual, tabular, and textual inputs. We compile these insights into a "Design Suggestion Deck" that surfaces the most influential layout features and proposes targeted optimizations. Experiments on the CircuitNet benchmark demonstrate that our approach outperforms existing models on both accuracy and explainability. Additionally, our design suggestion guidance case study and qualitative analyses confirm that the learned preferences align with real-world design principles and are actionable for engineers. This work highlights the potential of MLLMs as interactive assistants for interpretable and context-aware physical design optimization.

Yun-Da Tsai, Yu-Che Tsai, Bo-Wei Huang et al.

With the emerging trend of GPT models, we have established a framework called AutoML-GPT that integrates a comprehensive set of tools and libraries. This framework grants users access to a wide range of data preprocessing techniques, feature engineering methods, and model selection algorithms. Through a conversational interface, users can specify their requirements, constraints, and evaluation metrics. Throughout the process, AutoML-GPT employs advanced techniques for hyperparameter optimization and model selection, ensuring that the resulting model achieves optimal performance. The system effectively manages the complexity of the machine learning pipeline, guiding users towards the best choices without requiring deep domain knowledge. Through our experimental results on diverse datasets, we have demonstrated that AutoML-GPT significantly reduces the time and effort required for machine learning tasks. Its ability to leverage the vast knowledge encoded in large language models enables it to provide valuable insights, identify potential pitfalls, and suggest effective solutions to common challenges faced during model training.